Clinical Chemistry AACC Online Job Center
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Clinical Chemistry 45: 1181-1189, 1999;
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit an electronic Letter to
the Editor about this paper
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fox, M. P.
Right arrow Articles by Schneider, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fox, M. P.
Right arrow Articles by Schneider, E.
Related Collections
Right arrow Laboratory Management
Right arrow Proteomics and Protein Markers
(Clinical Chemistry. 1999;45:1181-1189.)
© 1999 American Association for Clinical Chemistry, Inc.

Articles

Effect of the Ratio of Free to Total Prostate-specific Antigen on Interassay Variability in Proficiency Test Samples

M. Pat Fox1, Andrew A. Reilly1,3 and Erasmus Schneider1,2,a

1 Division of Molecular Medicine, Wadsworth Center, New York State Department of Health, Albany, NY 12201.
Departments of
2 Biomedical Sciences and
3 Biometry and Statistics, School of Public Health, University at Albany, Albany, NY 12201.
a Address correspondence to this author at: State of New York Department of Health, Wadsworth Center, Empire State Plaza, P.O. Box 509, Albany, NY 12201. Fax 518-473-2900; e-mail schneid{at}wadsworth.org


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Background: Up to sevenfold differences were observed between total prostate-specific antigen (PSA) methods for New York State Proficiency Test samples prepared with seminal fluid PSA in human female serum. Because the PSA was mainly in its free form under these conditions, we wanted to determine whether a defined mixture of free and complexed PSA would reduce the interassay differences.

Methods: We prepared a series of five solutions of 60 g/L bovine serum albumin with 10 µg/L total PSA consisting of varied proportions of free, noncomplexible PSA, and {alpha}1-antichymotrypsin (ACT)-complexed PSA from 0% to 100%. Two hundred seventy laboratories measured the total PSA in these samples, and 16 laboratories also analyzed the samples for free PSA. The results were used to calculate free/total PSA ratios.

Results: Interassay CVs for total PSA measurements were ~7% at 10–15% free PSA but became gradually larger as the free/total PSA ratio increased. Measured free-PSA concentrations were similar within each sample (mean CV, 12%), and the results were relatively independent of the proportion of free PSA in the samples. Twofold discrepancies between actual and expected ratios were observed with some methods at 100% free PSA and to a lesser degree at 30% free PSA. At 100% free PSA, the relatively higher total-PSA values measured by nonequimolar methods yielded low free/total PSA ratios of 50–60%. In contrast, the lower total PSA values obtained by equimolar methods yielded ratios close to the expected 100%.

Conclusions: Preparing proficiency test samples with a 10:90 mixture of free, noncomplexible PSA:PSA-ACT is a viable alternative to the use of seminal fluid PSA. Furthermore, the method used to measure total PSA may have a substantial impact on the calculated proportion of free PSA and hence may have clinical relevance.© 1999 American Association for Clinical Chemistry


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Prostate-specific antigen (PSA)2 is a serine protease that is produced primarily, although not exclusively, in the prostate gland and is secreted into the seminal fluid (1)(2). PSA exists in at least three different molecular forms in human serum: free (uncomplexed) PSA (fPSA), PSA bound to {alpha}1-antichymotrypsin (PSA-ACT), and PSA bound to {alpha}2-macroglobulin. Of those, PSA-ACT is the predominant form (3)(4)(5), and only fPSA and PSA-ACT are detected by current immunoassays and therefore contribute to total-PSA measurements (3)(4)(5)(6)(7). Commercially available assays fall into two broad categories based on their relative ability to detect those two forms of PSA: equimolar assays or skewed-response (nonequimolar) assays (3)(5)(8)(9)(10). Equimolar assays measure both fPSA and PSA-ACT equally, and results obtained with these assays are largely independent of the relative amounts of fPSA and PSA-ACT in the test material. In contrast, skewed-response assays preferentially recognize the free form of PSA, thus measuring apparently higher total-PSA concentrations when the proportion of fPSA increases (11). Total-PSA concentrations reportedly increase in men with prostatic disease, and serum PSA concentrations are widely used in screening and disease management (10)(12). The relative amount of fPSA is higher in benign prostatic hyperplasia (BPH) compared with cancer of the prostate (CaP), and the ratio of free to total PSA may have diagnostic value for the distinction of BPH from CaP (13). Thus, differences in the response of various assays may be of clinical importance. It has been suggested that the higher molar response for fPSA by skewed-response assays might be reflected in higher total-PSA values for non-cancer patients who generally have higher proportions of fPSA compared with cancer patients (11)(14). This bias could potentially lead to false-positive results. This effect may be further exacerbated when the ratio of free to total PSA is calculated using results from nonequimolar assays. Because the higher total-PSA values are entered as the denominator in the calculation, the relative amount of fPSA calculated appears smaller and possibly suggests the presence of cancer in cases in which no malignant disease is present. Although this may be partially corrected for with assay-specific cutoff values (15), the issue of interassay variability and the relative role of varied ratios of fPSA and PSA-ACT in this variability may have some clinical relevance.

Several organizations, such as the College of American Pathologists and the State of New York, conduct proficiency testing (PT) as a means of external quality assessment of clinical laboratories. For PSA, PT samples usually are prepared by supplementing pooled human serum with partially purified seminal fluid PSA (16). Using samples prepared this way, a comparison of results obtained from the first New York State Tumor Marker PT event showed that there was significant interassay variability among the total-PSA concentrations measured with different commercially available kits, similar to what had been reported from a College of American Pathologists Basic Ligand Survey (16). In contrast, much smaller differences between methods were observed when semen-free serum samples served as the test material (16). This suggested that the discrepancies seen with PT samples were, at least partially, attributable to the nature of the test material. Several investigators had previously observed that discrepancies among different commercial assays varied with the ratio of fPSA to PSA-ACT in samples, especially at the more extreme ratios (7)(17)(18). Different values have resulted for the most part because measurements were made either with equimolar assays or with skewed-response assays (3). Therefore, in an attempt to produce PT samples with minimal interassay variabilities, test material was prepared with various ratios of fPSA to PSA-ACT to investigate to what extent the relative amount of fPSA contributed to the interassay variability observed.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
materials
Human serum from nonpregnant females was obtained from Intergen Company. Thirty percent bovine serum albumin (BSA) solution was obtained from either Sigma or Intergen. Standard-grade PSA (cat. no. 30-AP16) isolated from seminal fluid was obtained from Fitzgerald Industries. Free, noncomplexible PSA (cat. no. P5124) and PSA-ACT (cat. no. P0624) were obtained from Scripps Laboratories. Total-PSA Tandem-E1 and free-PSA Tandem-R assay kits were from Hybritech/Beckman. Assays in our laboratory were performed manually and in duplicate according to the manufacturer's instructions.

sample preparation
Standard-grade seminal fluid PSA was added in various amounts to human female serum. In separate samples, different proportions of free, noncomplexible PSA and PSA-ACT were added in various amounts to either human serum or 60 g/L BSA in phosphate-buffered saline containing 2 mmol/L EDTA (pH 7.3) as indicated in the respective results. All samples were sterile-filtered and kept liquid at all times. Aliquots were stored at 4, 22, and 37 °C, respectively, as indicated. All samples were initially analyzed in our laboratory immediately after preparation and at various times thereafter for total PSA by Tandem-E assay and for fPSA by Tandem-R assay. Some of the materials (in aseptically dispensed liquid aliquots) were also sent to clinical laboratories around the US as part of the New York State (NYS) tumor marker PT program. Those samples, in addition to PSA, also contained various amounts of {alpha}-fetoprotein, carcinoembryonic antigen, and CA125. None of these antigens interfered with PSA measurements (data not shown).

statistical analysis
For the stability studies (Fig. 1 , B-E), linear models in time were fit by classical least squares. Tests for parallelism and coincidence were conducted with sequential F-tests, utilizing the appropriate degrees of freedom (19). For the method comparison studies ( Figs. 2–5 ), analysis of covariance was used to account for the differing sample concentrations and/or percentage of fPSA. Both method target and slope differences were assessed with multiple comparison F-tests. Consequently, significance levels were taken as 0.001 to protect against false null hypothesis rejections (20). The error bars represent one side of the two-sided 95% confidence intervals computed from estimated variances and depend on the values of the covariates. Relative values were produced by normalizing raw values with the precision weighted means from all laboratories. All run programming was done in APL (21).



View larger version (25K):
[in this window]
[in a new window]
  		Figure 1. Stability of various PT sample preparations.

(A), standard-grade seminal fluid PSA was added at various concentrations to human female serum and stored at 4 °C ({blacksquare}), 20 °C ({diamondsuit}), or 37 °C (•). At various times thereafter, total PSA was measured. (B and C), a mixture of 10% free, noncomplexible PSA and 90% PSA-ACT was added at various concentrations to either human female serum (B) or 60 g/L BSA in PBS, pH 7.3 (C) and stored at 4 °C. At various times thereafter, total PSA was measured. (D), fPSA was measured in samples from B ({circ}, serum) and C (•, BSA). (E), the ratio of free to total PSA was calculated in samples from B ({circ}, serum) and C (•, BSA). For A, all data points are the means of the measurements from three different concentrations and are expressed as the percentage remaining with standard errors compared to the first measurement made immediately after preparation. For B–E, all data points from multiple individual measurements are given and were used for linear regression analysis. Lines shown are for linear regression with 95% confidence intervals. For samples in BSA, results from several independent preparations were combined. Total PSA was measured with the Hybritech Tandem E, and fPSA was measured with the Hybritech Tandem R.



View larger version (22K):
[in this window]
[in a new window]
  		Figure 2. Relative total-PSA concentrations.

Standard-grade seminal fluid PSA was added to human female serum at nominally 10, 50, and 200 µg/L. Samples were sent to NYS-permitted laboratories for analysis (n = 240). Results were analyzed by method peer groups as indicated and normalized to the average from the weighted means of all methods after accounting for differences in sample concentrations. Error bars represent 95% confidence intervals. Intermethod comparison permits (P = 0.08) grouping of similar methods as follows: Abbott AxSym, Abbott IMx, and Chiron PSA2; Bayer Immuno-1 and DPC; Tosoh, Hybritech Tandem E, and Hybritech Tandem R. Chiron PSA 1 may not be assigned to any other group (P <0.00005). Further coalescing is not possible.



View larger version (26K):
[in this window]
[in a new window]
  		Figure 3. Relative total-PSA concentrations.

Mixtures of free, noncomplexible PSA and PSA-ACT to give ~10% fPSA were added at various total concentrations to either human female serum (A) or 60 g/L BSA in PBS, pH 7.3 (B). Samples were sent to NYS-permitted laboratories for analysis (n = 260). Results were analyzed by method peer groups as indicated and normalized to the weighted average from the means of all methods after accounting for differences in sample concentrations. Error bars represent 95% confidence intervals. Measured/calculated free/total PSA ratios were 11.4% (A) and 13.1% (B), respectively. With the exception of Chiron PSA1, differences among the different methods are not statistically significant (P >0.05).



View larger version (36K):
[in this window]
[in a new window]
  		Figure 4. Total PSA as a function of fPSA.

Five samples of 10 µg/L total PSA each were prepared in 60 g/L BSA with 0%, 5%, 15%, 30%, and 100% free, noncomplexible PSA. Samples were sent to NYS-permitted laboratories for analysis (n = 270). Results were analyzed by method peer groups as indicated. Error bars represent 95% confidence intervals. (A), total-PSA method comparison within each sample. The inset represents the ratio between highest and lowest measured total PSA, excluding results obtained with the Chiron PSA1 assay. Columns represent weighted means of results obtained with (from left to right): Abbott AxSym, Abbott IMx, Bayer Immuno-1, Chiron PSA1, Chiron PSA2, DPC Coat-a-Count, DPC Immulite, Hybritech Tandem-E, Hybritech Tandem-R, and Tosoh. (B), results for each method as a function of the percentage of fPSA. {blacksquare}, Chiron PSA2; {blacktriangleup}, Abbott AxSym and IMx combined; {blacktriangledown}, DPC Immulite; {diamondsuit}, Bayer Immuno-1; , Hybritech Tandem E and R combined; •, Tosoh.



View larger version (39K):
[in this window]
[in a new window]
  		Figure 5. fPSA concentrations.

(A), fPSA method comparison in samples from Fig. 4Up . Columns represent means of results obtained with (from left to right): Abbott (n = 3; CV, 5.6–16%), DPC (n = 4; CV, 5.1–15%), and Hybritech (n = 6; CV, 4.5–14%) methods, respectively. Error bars represent 95% confidence intervals. (B), calculated percentage of fPSA. Columns represent the calculated percentage of fPSA, using the method combinations (free/total) as follows (from left to right): Abbott/Abbott AxSym (n = 3), DPC/Abbott IMx (n = 1) or AxSym (n = 1), Hybritech/Hybritech Tandem E (n = 2) or Tandem R (n = 4), DPC/Chiron PSA2 (n = 1), and DPC/Bayer (n = 1). Note the change in the y-axis scale for the 100% sample.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
stability
The stability over time of standard-grade seminal fluid PSA in human female serum is shown in Fig. 1Up A. Not surprisingly, the stability of PSA was both time and temperature dependent. At 4 °C, PSA concentrations gradually decreased over 23 days to ~80% of the initial PSA value measured, at 22 °C they decreased to ~60%, and at 37 °C they decreased to ~40%. Thereafter, when samples were kept at 4 °C, PSA concentrations stabilized with only a small further loss of 5–10% over the next 5–10 weeks, whereas at higher temperatures the loss of PSA continued over time. Thus, it appears that after an initial period of "equilibration", which is accompanied by a loss of 15–20% of PSA at 4 °C, the concentrations stabilized and remained relatively constant for at least 1 month. In contrast, when samples were prepared with a mixture of free, noncomplexible PSA and PSA-ACT at various ratios of 10–20% fPSA, total-PSA concentrations remained fairly constant for as long as 9 months when stored at 4 °C (Fig. 1Up , B and C).

Total PSA stability in serum was different from that in BSA (P = 0.0001), with the concentration in BSA remaining essentially constant (P = 0.134 for slope != 0) and the concentration in serum decreasing by ~1.5% per month (P <0.00005). In addition to total PSA, we also determined the stability of the fPSA component in some of these samples. As shown in Fig. 1DUp , in BSA there was a slow but gradual increase of 1.5% per week in the amount of fPSA measured (P <0.00005), whereas it remained essentially stable in serum (P = 0.38 for slope != 0). The difference in fPSA stability between serum and BSA was statistically significant (P <0.00005). When we used the data from Fig. 1Up , B-D, to calculate the ratios of free to total PSA, there was an increase over time that was not significantly different between BSA and serum (P = 0.004; Fig. 1EUp ). Although we have not further investigated the cause for this apparent increase in fPSA over time, possible reasons include a gradual decay of the PSA-ACT complex, similar to the result reported by Pettersson et al. (6).

In conclusion, these studies demonstrate that it is possible to prepare PT samples for PSA with a mixture of free, noncomplexible PSA and PSA-ACT that are stable over a period of several months. This time is sufficiently long to prepare, quality control, ship, and analyze the material without concern for result bias as a result of different times to analysis in the participating laboratories.

method comparison: total psa
In May 1996, five samples were sent to each of 240 laboratories that participated in the tumor marker PT for NYS. The test material was human female nonpregnant serum that had been supplemented with 0, 4, 10, 50, and 200 µg/L standard-grade seminal fluid PSA. (Determination of the nominal concentrations of PSA was based on the concentration of the material used to supplement the matrix, as provided by the manufacturer. Samples with 0 and 4 µg/L PSA were not included in the following analysis because of insufficient recovery).

As can be seen in Fig. 2Up , there were large differences in measured PSA with various assay kits (interassay CV, 55%). The measured PSA concentrations ranged from a low of 0.30 to a high of 1.82 (mean, 0.88 ± 0.49; median, 0.69) times the weighted average of the means from all methods, reflecting a sixfold difference from lowest to highest. Measured PSA concentrations generally were lower than expected, which may, at least in part, be attributed to the loss of PSA during the initial equilibration period described above (samples were sent to individual laboratories 25 days after preparation). However, it seemed unlikely that this could explain the large interassay differences observed. Rather, the results suggested that the differences obtained with different methods were related to the nature of the PSA in the samples with >80% of the PSA in its free form [data not shown, see also Ref. (22)]. In addition, it was possible that the different assay characteristics such as poly/monoclonal vs mono/monoclonal antibodies and assay architecture or reaction kinetics may also have affected the performance of an individual assay (23). Although not unique (16)(17)(18), it was nevertheless unsatisfactory when comparing results for PT purposes. Therefore, in an attempt to obtain a more uniform response from the different methods, the next two sets of NYS PT samples were prepared using a mixture of ~10% free, noncomplexible PSA and 90% PSA-ACT in serum, based on a recommendation by Stamey et al. (24) and Chen et al. (25). As shown in Fig. 3Up A, there was good correlation in these samples between results obtained with different assay methods at various PSA concentrations from 2.5 to 25 µg/L, with the exception of the original PSA1 assay by Ciba-Corning (now Chiron). Values obtained with individual assays differed from the average of the means (excluding Chiron PSA1) from all assays by <10% (interassay CV, 8.3%). Similar results were obtained when samples were made in a matrix of 60 g/L BSA in phosphate-buffered saline instead of serum (Fig. 3BUp ; interassay CV, 7.7%). Thus, the use of a mixture of free, noncomplexible PSA and complexed PSA appeared to significantly reduce or even eliminate interassay differences.

Because using a mixture of ~10% free, noncomplexible PSA and 90% PSA-ACT seemed to largely eliminate differences for total-PSA measurement between most assays, we next investigated the role of various amounts of free, noncomplexible PSA on the relative performance of different assay methods. For this investigation, PT samples were made with 0–100% free, noncomplexible PSA complemented with appropriate amounts of PSA-ACT to achieve constant concentrations of 10 µg/L total PSA as indicated in Fig. 4Up A. As can be seen from these data, the total-PSA concentrations measured by the majority of the different assays were fairly equal up to 15% fPSA (minimum–maximum range, 1.2-fold, excluding Chiron PSA1; see inset in Fig. 4AUp ), whereas a divergence between the different assays became apparent at higher proportions of fPSA. As the amount of fPSA became proportionally larger, the range among the different assay methods increased up to almost threefold at 100% fPSA. In Fig. 4BUp , values for total PSA measured with different methods were graphed against the proportion of fPSA in each sample. These data clearly show that the proportion of PSA present in its free, uncomplexed form can have a rather major impact on the overall total-PSA concentration measured with different methods.

Of the six manufacturers compared, increasing fPSA produced a slight downward trend with assays from three manufacturers, whereas a more or less steep upward trend was observed with assays from the other three. Only one method (Bayer) measured total-PSA values at 100% fPSA that were within 20% of those measured at 0% fPSA (ratio, 0.89; P = 0.001 for slope != 0). This suggested that at least in this sample set, this assay was the most equimolar in its behavior. Of the other assays, those from Hybritech and Tosoh appeared to exhibit some bias for the complexed form of PSA, whereas the assays from Abbott and Chiron, and to a lesser degree the assay from Diagnostics Products Corp. (DPC), showed a clear bias for the free form of PSA (P <0.00005 for slope != 0). Interestingly, the two least equimolar assays measured somewhat lower values than the rest at 0% fPSA, with the "crossover" to higher values occurring at 3–5% for the Chiron assay and 15–20% for the Abbott assay. Since completion of this study, similar results were published by Cheli et al. (17) and Blase et al. (18).

method comparison: fPSA
While the experiments described above were underway, several methods to measure fPSA became available. Therefore, laboratories participating in the NYS PT program were asked to also measure fPSA in the samples provided; 16 laboratories did so (n = 16). As shown in Fig. 5Up A, of the three assays used by more than one laboratory (three laboratories used an assay not used by anyone else), one showed a small bias for higher values, especially at high relative concentrations of fPSA. fPSA concentrations measured by the three assays were similar within each sample (mean CV, 12%; range, 8.5–16%; P = 0.01), and the results were relatively independent of the proportion of fPSA in the samples, at least within the diagnostically relevant range of 0–30% fPSA. Essentially the same results were obtained when we repeated this experiment more recently with more participating laboratories (n = 27; mean CV, 10%; range 8.0–13%).

calculating percentage of fPSA
Because the measurements for free and total PSA described above were made on the same set of samples, we also examined how the choice of assays would influence the calculation of the percentage of fPSA, which is the clinically relevant indicator. From the small subset of laboratories that had reported values both for total and free PSA we calculated the proportions of fPSA in each sample and compared them with what had been added to the samples. Fig. 5BUp shows that when the sample contained only fPSA, the results fell into two groups that differed by up to a factor of two. Although only free, noncomplexible PSA had been added to this sample, three of the combinations led to calculated values of between 50% and 60%, whereas the other two gave the expected result of ~100% fPSA. Qualitatively similar, although less pronounced, results were also obtained at lower proportions of fPSA (32% vs 20%, calculated in samples that had a 30:70 fPSA:PSA-ACT mixture added). The difference between these two groups was highly significant (P = 0.0004). Not surprisingly, combinations that used a total-PSA assay with a bias for fPSA led to lower calculations of the percentage of fPSA. Comparison of the amount of total PSA with that of fPSA measured, which at 100% fPSA should be identical, indicated that the nonequimolar-response total-PSA assays apparently measured the noncomplexible PSA twice as efficiently as fPSA assays did. This result is similar to those reported previously by Cheli et al. (17) and Blase et al. (18). Interestingly, this also appears to be true in patient samples (26). Thus, the analytical behavior of our samples appeared to mimic that of patient samples.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
PT is used by various professional and regulatory organizations for external quality control of clinical laboratories. At the same time, it gives those laboratories a chance to compare their performance to that of their peers, despite the caveat that the test materials distributed usually are artificial and may not always reflect true patient samples. Unfortunately, those comparisons often are additionally hindered by a lack of standardization of both the test material and the tests themselves. This became evident when results from the first NYS PT event for PSA were compared. Differences of up to sevenfold between the lowest and highest results were reported, although differences between results obtained with identical methods/kits were much smaller (peer group CV <10–15% after purging of outliers). This was rather unsatisfactory and, following a suggestion by Stamey et al. (24) and Chen et al. (25), we therefore decided to prepare PSA PT test material with a 10:90 mixture of free, noncomplexible PSA and ACT-complexed PSA. Subsequent comparison of NYS PT results from almost 250 laboratories showed a clear improvement in interassay variability (all-laboratory CV <10%) with the exception of the original Chiron PSA1 assay. This assay consistently gave substantially higher results at all concentrations of PSA. Because part of the variability seen originally may have resulted from an interaction of the human serum matrix with the PSA material added to the samples, the test material was further modified by use of a 60 g/L BSA solution as the matrix. It is evident from the results presented that PSA PT material prepared in this way leads to a tight correlation between most assays and is therefore much better suited for interlaboratory comparisons. This material is also very stable, with negligible loss of reactivity for up to 9 months when stored at 4 °C. Thus, samples do not need to be lyophilized, thereby eliminating potential errors from reconstitution in the tested laboratories. Furthermore, assays for total, free, and complexed PSA can be evaluated in the same sample and, although not presented in this report, we have also found that additional tumor markers such as {alpha}-fetoprotein, carcinoembryonic antigen, CA125, CA15-3, and CA19-9 can be added without noticeable effects on performance. Thus, we believe that PT material for tumor markers prepared in BSA is a valid and probably better controlled alternative to the human serum usually used. Although the data presented in the present report are based on NYS PTs in 1996 and 1997, we have now used material that was prepared in this manner for several more PT events and obtained essentially identical results.

Since the present studies were completed, several new assays have been introduced in the US. Several of those showed significantly different results from the mean (data not shown) despite the otherwise good interassay correlation with the new material. For example, results reported with the Boehringer Mannheim Roche Elecsys were consistently 20–30% lower than the mean, and this seemed independent of the proportion of fPSA present, whereas those obtained with Beckman Access were consistently higher (data not shown). Similar observations were reported recently by Stamey et al. (27) in patient-derived material. However, it is unlikely that test material can be made to "accommodate" all assays. Although the number of laboratories using these newer assays in the NYS surveys is still too small to allow a definitive conclusion, it seems that additional standardization has to come from adapting or recalibrating some of the assays. This has happened relatively successfully with the development of the Chiron PSA2 assay (28)(29) and was also successfully demonstrated in the study by Stamey et al. (27) for the Boehringer Mannheim Roche Elecsys assay. The availability of standardized material as recently recommended by NCCLS should aid manufacturers in this endeavor (30).

In light of the recent trend to use the ratio of free to total PSA as an indicator to aid in the discrimination of CaP from other types of prostatic disease, especially BPH, a closer examination of the factors involved in the determination of this ratio was thought to be useful. To this effect PT samples with various ratios of free, noncomplexible PSA and PSA-ACT were prepared and sent to NYS-permitted laboratories for analysis. Although measurements of both free and total PSA were made by a relatively small subset of laboratories (n = 16; this number has almost doubled in more recent PT events without substantially changing the conclusions from this report), they were made with several different kits that were commercially available for the analysis of these analytes. The overall consensus between laboratories for fPSA in our samples was substantially better than that reported from a similar study by Zuchelli et al. (15). In that report from the European Extended Quality Assessment Program, the mean between-laboratory agreement of fPSA determinations had a CV of 28%, and the authors concluded that the "average fPSA results produced by the three most popular methods are consistently different from each other". It is not immediately clear why the results from their study were different from those presented here. One possibility is the different nature of the samples. Whereas we added a mixture of purified free, noncomplexible PSA and PSA-ACT to a nonhuman protein matrix, the Extended Quality Assessment Program Oncocheck samples were prepared by diluting a serum pool obtained from cancer patients with normal human serum. When we calculated the ratio of free to total PSA, we observed that in general, using method combinations where the assay for total PSA has been reported to be equimolar produced a ratio close to what was expected. In contrast, combinations with a total-PSA assay that has been reported to be nonequimolar produced lower than expected ratios of free to total PSA. This discrepancy is not surprising because values for total PSA concentration are entered as the denominator in the calculations. Various reports have discussed the validity and utility of using the free-to-total PSA ratio to help to distinguish between CaP and BPH to allow the early detection of CaP while limiting the number of negative and therefore unnecessary biopsies (31)(32). The results reported here suggest that the methods used to determine the free/total PSA ratio may have a major impact and should, therefore, be carefully considered in discussions concerning the cutoff values used in making clinical decisions (15). Although the differences in the free/total PSA ratios determined with some kits were almost twofold at 100% fPSA, which is not likely to be a clinically relevant proportion occurring in the natural patient population (33), our data suggest that in the more relevant range of 25–40%, the differences might be enough to push a decision for or against biopsy if one type of method combination with a bias was used vs another without bias (11). At 30% fPSA, the calculated percentage of fPSA with the biased methods was approximately one-quarter to one-third lower than that calculated with the latter (20% free vs 32% free). Thus, our results suggest that any cutoff recommendations for free/total PSA should take into account not only the method for fPSA, but also and more importantly, the assay used for total PSA measurement (15)(34).

In conclusion, we have shown that artificial PT material for PSA can be produced that is both stable and has minimal interassay differences, at least with the major methods currently used in the US. We believe that this material meets the criterion for commutability as defined by Rej (35) because it appears to mimic the response of various assays in patient samples. We have furthermore shown that different assays respond differently to variations in the ratio of free to complexed PSA in test samples and that those differences may possibly affect the clinical sensitivity of this assay.


   Acknowledgments
 
We thank the laboratories that participate in the NYS PT program and whose results were used for the present study. We also thank the Computational Molecular Biology and Statistics Core of the Wadsworth Center for the use of hardware and software.


   Footnotes
 
Portions of this work were presented previously in abstract form in Clinical Chemistry 1998;44:A41.

2 Nonstandard abbreviations: PSA, prostate-specific antigen; fPSA, free form of PSA; PSA-ACT: {alpha}1-antichymotrypsin-complexed PSA; BPH, benign prostatic hyperplasia; CaP, cancer of the prostate; PT, proficiency test; BSA, bovine serum albumin; NYS, New York State; and DPC, Diagnostics Products Corp.

1 The use of brand and/or trade names in this report does not constitute an endorsement of the products on the part of the Wadsworth Center or the New York State Department of Health.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Wang MC, Papsidero LD, Kuryiama M, Valenzuela LA, Murphy GP, Chu TM. Prostate antigen: a new potential marker for prostate cancer. Prostate 1981;2:89-96. [ISI][Medline] [Order article via Infotrieve]
  2. Lilja H, Abrahamsson PA. Three predominant proteins secreted by the human prostate gland. Prostate 1988;12:29-38. [ISI][Medline] [Order article via Infotrieve]
  3. McCormack RT, Rittenhouse HG, Finlay JA, Sokoloff RL, Wang TJ, Wolfert RL, et al. Molecular forms of prostate specific antigen and the human kallikrein gene family: a new era. Urology 1995;45:729-745. [ISI][Medline] [Order article via Infotrieve]
  4. Lilja H, Christensson A, Dahlén U, Matikainen MY, Nilson O, Pettersson K, et al. Prostate-specific antigen in serum occurs predominantly in complex with {alpha}1-antichymotrypsin. Clin Chem 1991;37:1618-1625. [Abstract/Free Full Text]
  5. Stenman U-H, Leinonen J, Alfthan H, Rannikko S, Tuhkanen K, Alfthan O. A complex between prostate-specific antigen and {alpha}1-antichymotrypsin is the major form of prostate-specific antigen in serum of patients with prostatic cancer: assay of the complex improves clinical sensitivity for cancer. Cancer Res 1991;51:222-226. [Abstract/Free Full Text]
  6. Pettersson K, Piironen T, Seppalla M, Liukkonen L, Christensson A, Matikainen M-T, et al. Free and complexed prostate-specific antigen (PSA): in vitro stability, epitope map, and development of immunofluorometric assays for specific and sensitive detection of free PSA and PSA-{alpha}1-antichymotrypsin complex. Clin Chem 1995;41:1480-1488. [Abstract/Free Full Text]
  7. Zhou AM, Tewari PC, Bluestein BI, Caldwell GW, Larsen FL. Multiple forms of prostate-specific antigen in serum: differences in immunorecognition by monoclonal and polyclonal assays. Clin Chem 1993;39:2483-2491. [Abstract]
  8. Graves HCB. Standardization of immunoassays for PSA. Cancer 1993;72:3141-3144. [ISI][Medline] [Order article via Infotrieve]
  9. Stamey TA, Chen Z, Prestigiacomo A. Serum prostate specific antigen binding {alpha}1-antichymotrypsin: influence of cancer volume, location and therapeutic selection of resistant clones. J Urol 1994;152:1510-1514. [ISI][Medline] [Order article via Infotrieve]
  10. Becker C, Lilja H. Individual prostate-specific antigen (PSA) forms as prostate tumor markers. Clin Chim Acta 1997;257:117-132. [ISI][Medline] [Order article via Infotrieve]
  11. Jacobsen SJ, Lilja H, Klee GG, Wright GL, Pettersson K, Oesterling JE. Comparability of the Tandem-R and IMx assays for the measurement of serum PSA. Urology 1994;44:512-518. [ISI][Medline] [Order article via Infotrieve]
  12. Chu TM. Prostate-specific antigen and early detection of prostate cancer. Tumor Biol 1997;18:123-134.
  13. Catalona WJ. Clinical utility of measurements of free and total prostate-specific antigen (PSA): a review. Prostate Suppl 1996;7:64-69. [Medline] [Order article via Infotrieve]
  14. Semjonow A, Oberpenning F, Brandt B, Zechel C, Brandau W, Hertle L. Impact of free prostate-specific antigen on discordant measurement results of assays for total prostate-specific antigen. Urology 1996;48:10-15. [ISI][Medline] [Order article via Infotrieve]
  15. Zuchelli GC, Pilo A, Chiesa MR, Cohen R, Bizollon A. Analytical performance of free PSA immunoassays: results from an interlaboratory survey. Clin Chem 1997;43:2426-2428. [Free Full Text]
  16. Garg UC, Howanitz JH, Nakamura RM, Plous RH, Eckfeldt JH. Production, analysis, and characterization of reference materials for PSA. Arch Pathol Lab Med 1995;119:1104-1108. [ISI][Medline] [Order article via Infotrieve]
  17. Cheli CD, Marcus M, Levine J, Zhou Z, Anderson PH, Bankson DD, et al. Variation in the quantitation of prostate-specific antigen in reference material: differences in commercial immunoassays. Clin Chem 1998;44:1551-1553. [Free Full Text]
  18. Blase AB, Sokoloff RL, Smith KM. Five PSA methods compared by assaying samples with defined PSA ratios. Clin Chem 1997;43:843-845. [Free Full Text]
  19. Ostle B, Mensing RW. Statistics in research: basic concepts and techniques for research workers 1975:210-214 Iowa State University Press Ames, Iowa. .
  20. Seber GAF. Linear regression analysis 1977:125-130 John Wiley & Sons New York. .
  21. STSC. APL Plus II system for Unix, Ver. 4 ed. Princeton, NJ: APL2000, 1991..
  22. Chen Z, Komatsu K, Prestigiacomo A, Stamey TA. Addition of purified prostate specific antigen to serum from female subjects: studies on the relative inhibition by {alpha}2-macroglobulin and {alpha}1-antichymotrypsin. J Urol 1996;156:1357-1363. [ISI][Medline] [Order article via Infotrieve]
  23. Tewari PC, Williams JS. Analytical characteristics of seminal fluid PSA differ from those of serum PSA. Clin Chem 1998;44:191-192. [Free Full Text]
  24. Stamey TN, Prestigiacomo AF, Chen Z. Standardization of immunoassays for PSA. Cancer 1994;74:1662-1666. [ISI][Medline] [Order article via Infotrieve]
  25. Chen Z, Prestigiacomo A, Stamey TA. Purification and characterization of PSA complexed to antichymotrypsin: potential reference material for international standardization of PSA immunoassays. Clin Chem 1995;41:1273-1282. [Abstract/Free Full Text]
  26. Blase AB, Sokoloff RL. Analytical characteristics of seminal fluid PSA differ from those of serum PSA [Reply to Letter]. Clin Chem 1998;44:192-193.
  27. Stamey TA, Chen Z, Prestigiacomo AF. Reference material for PSA: the IFCC standardization study. International Federation of Clinical Chemistry. Clin Biochem 1998;31:475-481. [ISI][Medline] [Order article via Infotrieve]
  28. Bankson DD, Petteway JC, Smith AA, Eisenreich CG, Green RE, Haver VM, et al. A recalibrated CIBA-Corning ACS-180 PSA (PSA2) assay: comparison with the Hybritech Tandem-R and Abbott IMx PSA assays [Abstract]. Clin Chem 1996;42:S266.
  29. Tewari P, Farrington K, Keelan M, Christensen S, Comerci C, Bluestein B, et al. ACS PSA2, a new immunoassay for the measurement of prostate-specific antigen (PSA) in serum [Abstract]. Clin Chem 1996;42:S265.
  30. NCCLS. Primary reference preparations used to standardize calibration of immunochemical assays for serum prostate-specific antigen (PSA); approved guideline. NCCLS document I/LA19-A. Wayne, PA: NCCLS, 1997..
  31. Filella X, Alcover J, Molina R, Rodriguez A, Carretero P, Ballesta AM. Free and total PSA in the diagnosis of prostate cancer. Tumor Biol 1997;18:332-340.
  32. Morote J, Raventos CX, Lorente JA, Lopez-Pacios MA, Encabo G, de Torres I, et al. Measurement of free PSA in the diagnosis and staging of prostate cancer. Int J Cancer 1997;71:756-759. [ISI][Medline] [Order article via Infotrieve]
  33. Catalona WJ, Smith DS, Wolfert RL, Wang TJ, Rittenhouse HG, Ratliff TL, et al. Evaluation of percentage of free serum prostate-specific antigen to improve specificity of prostate cancer screening. JAMA 1995;274:1214-1220. [Abstract]
  34. Junker R, Brandt B, Zechel C, Assman G. Comparison of prostate-specific antigen (PSA) measured by four combinations of free PSA and total PSA assays. Clin Chem 1997;43:1588-1594. [Abstract/Free Full Text]
  35. Rej R. Accurate enzyme activity measurements: two decades of development in the commutability of enzyme quality control materials. Arch Pathol Lab Med 1993;117:352-364. [ISI][Medline] [Order article via Infotrieve]



The following articles in journals at HighWire Press have cited this article:


Home page
 Clin. Chem.Home page
C. Stephan, H. Cammann, A. Semjonow, E. P. Diamandis, L. F.A. Wymenga, M. Lein, P. Sinha, S. A. Loening, and K. Jung
Multicenter Evaluation of an Artificial Neural Network to Increase the Prostate Cancer Detection Rate and Reduce Unnecessary Biopsies
Clin. Chem., August 1, 2002; 48(8): 1279 - 1287.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Chem.Home page
D. W. Chan and L. J. Sokoll
WHO First International Standards for Prostate-specific Antigen: The Beginning of the End for Assay Discrepancies?
Clin. Chem., September 1, 2000; 46(9): 1291 - 1292.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit an electronic Letter to
the Editor about this paper
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fox, M. P.
Right arrow Articles by Schneider, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fox, M. P.
Right arrow Articles by Schneider, E.
Related Collections
Right arrow Laboratory Management
Right arrow Proteomics and Protein Markers

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS